Lake Effect Snow Forecast Tools at NWS Gaylord, MI Part 1: The Similar Sounding Approach Justin Arnott Science and Operations Officer NWS Gaylord, MI

Introduction “Legacy Approach” ◦ Threat Maps ◦ Flowcharts ◦ 12 km NAM ◦ Pattern Recognition ◦ BUFKIT Demand for improved spatial/temporal evolution in the NDFD Can we improve upon these tools? ◦ LE Similar Sounding approach (this talk) ◦ High Resolution Local WRF (Part II)

The BGM Approach (Evans and Murphy 2008) Archive BUFKIT sounding graphics for lake effect events ◦ One “representative sounding” per event. Derive LES-important parameters from these soundings ◦ Create database Compare current forecasts to database to find the most “similar soundings” ◦ Retrieve radar imagery/summary of similar events

The APX Approach Archive RUC BUFR soundings every hour for the winter season Derive sounding-based parameters ◦ Similar to those at BGM with some exceptions  ML Wind Direction/Speed  Inversion Height/Delta T  850 T/Delta T  Moist Depth  Mean Tdd (Below 900, , )  DGZ Depth  Bulk Shear

The APX Approach - Continued Create CWA-mean values ◦ Use apx,kpln,anj,lm1,ktvc,kcad,kciu BUFR sites Create database of RUC-Based Parameters “Match” current forecast soundings with database to find similar events ◦ Scoring a blend of objectivity/subjectivity ◦ Three ‘tiers’ of scoring

The APX Approach - Continued Provide 1 hour loop of radar from matching events ◦ Use NCDC Level 3 archive ◦ More information than snapshot?

The APX Approach – To Date Period of Archive ◦ 2010NOV FEB22 ◦ 57 days with lake effect ◦ 1127 ‘sounding hours’ Removed hours with significant synoptic precipitation

The Result Full Weight Half Weight Quarter Weight Run Information

One Hour Base Reflectivity Loop From 12/23/

One Hour Base Reflectivity Loop From 12/23/

Caveats RUC 00hr Sounding = Truth? ◦ RUC is non-hydrostatic with relatively few vertical levels – but runs every hour! ◦ Will be replaced with Rapid Refresh once it becomes operational Many parameters not yet taken into account ◦ Time of Day/Time of Year ◦ Persistence ◦ A plethora of meteorological-parameters

Let’s Turn This Approach Around Have netCDF Level 3 base reflectivity Can we look for sounding parameters that correlate to different reflectivity patterns? ◦ E.G. What parameters bring high reflectivities far inland? ◦ What parameters bring heavy snow to Gaylord? Alpena, Traverse City, etc.

Site Specific “LES Roses” Data: ◦ Hourly average point reflectivity data  For this presentation use KGLR/KTVC ◦ Hourly Mixed Layer Mean Wind direction

~52 total hours

Why the Gap?

KGLR degree maximum ◦ Confirms conceptual models

~61 total hours

KTVC degree maximum

Other Investigations? Variable scatterplots

Summary APX Similar Sounding Approach ◦ Similar to BGM, yet different!  Hourly RUC Soundings  Average of numerous CWA BUFR points ◦ Now operational to forecasters ◦ Over 1000 soundings in one cool season ◦ Should improve given more data

Summary – continued Using Archived LES Radar data ◦ “LES Roses” ◦ Variable/Reflectivity scatterplots ◦ New forecaster training tool  How does LES behave in the APX CWA.  When does TVC get hit? GLR?

Future Plans New Sounding-based variables ◦ Surface to inversion base mean Tdd, Directional shear, etc. Move away from average approach ◦ Have a score for EACH sounding site, then make a composite score. Reassess weights for creating “matches” Add COOP data in addition to radar ◦ Current approach limited for eastern U.P. Move from 2D to 3D approach? ◦ E.G. Like a LES-tuned CIPS?

Future Plans - Continued Develop reflectivity-based LES climatology ◦ Create LES Roses for multiple sites ◦ Deeper investigations of parameter/reflectivity correlations  Use to adjust similar sounding matching techinque  Move away from just point-based  Banded vs. cellular  Inland extent ◦ Expand Training Tool Potential  Intro to LES  Debunk LES Myths

Acknowledgements NCDC ◦ For radar data and Weather and Climate Toolkit export capability APX Forecasters ◦ For testing tool